DOCSLIB.ORG

Probability: What Affects Estimates

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Probability: What Affects Estimates Spring 2010 Math 263 Deb Hughes Hallett Class 13: Confidence Intervals for Means Statistical Inference We take a sample to learn about a population. There are two ways that we can draw a conclusion: Estimation, using confidence intervals. Here we use the sample to make an estimate of a population parameter, such as the population, ,.or the population proportion, . --For example, estimate the mean income in a community from a sample. Hypothesis testing. Here we test a claim about a population. --For example, test the claim that a drug lowers blood pressure significantly. Example: What is the Effect of the Police Radar US traffic police often use radar to catch drivers speeding. To alert them to the presence of police radar, some drivers mount radar detectors in their cars. This has led to a debate:1 Are radar detectors a useful reminder to stay within the speed limit, or are they simply a way of avoiding police detection? A study2 in Maryland found that a sample of 22 cars with radar detectors slowed down an average of 11 mph in the presence of radar. Suppose that the speed reduction of individual cars was normally distributed with standard deviation 2 mph.3 Ex: What does this sample tell us about the average drop in speed of all cars with radar detectors? What is: Variable type: (Quantitative/categorical?): Quantitative Population: All cars with radar detectors Population Parameter: Average drop in speed of all cars with radar detectors Sample: The 22 cars sampled Sample Statistic: Average drop in speed of cars in sample, 11 mph Estimate of population mean: We use the sample mean, 11 mph, as an estimate of the population mean. How far from the true mean could this estimate be? Confidence Intervals To see how far from the true value our estimate of 11 mph could be, we construct a confidence interval, in which the true population mean is likely to lie. The margin of error and the width of the confidence interval depend on how much the sample means vary between samples; this is determined by the Central Limit Theorem.4 The Central Limit Theorem tells us the mean drop in speed for a sample of 22 cars is normally distributed with mean equal to the mean drop in speed of the population (which we don’t know) and standard deviation = mph. Suppose the mean drop in speed for the population was 11 mph. (Note: It wasn’t exactly 11 mph, as 11 mph is the sample mean, but we expect the population mean is close to 11 mph.) Then the distribution of sample means for samples of size 22 would look like this: 1 From Ohio State’s EESEE, based on work by N.Teed, K.Adrian, R. Khoblanch, 1991, www.whfreeman.com/scc6e 2 www.afn.org/nafn 09444/ scanlaws/ 3 We are going to need to know the standard deviation of the population distribution, so we take this to be 2. 4 We can use the Central Limit Theorem even though the sample size is less than 30 because the original distribution is normal. 1 Spring 2010 Math 263 Deb Hughes Hallett Distribution of Average Drop in Speed for Samples of 22 Cars Mean 11 mph, Standard deviation 2 mph 0.0 9.00 10.00 11.00 12.00 13.00 Drop in speed (mph) The graph suggests almost all the mean drops in speed are between 10 mph and 12 mph. Since 95 % of the data is within 2 standard deviations of the mean, we conclude that 95% of the drops in speed are roughly between 11 – 2 (0.43) mph and 11 + 2 (0.43) mph = 11 – 0.86 mph and 11 + 0.86 mph = 10.14 mph and 11.86 mph. The interval is called a confidence interval. More accurate Confidence Interval Ex: Use the table to find a more accurate the -values on either side of 0 containing 95% of the data. We want the z-values leaving 2.5% on the outside; the closest value is and More precisely, we can now say that 95 % of the speed drops are between 11 – 1.96 (0.43) mph and 11 + 1.96 (0.43) mph = 11 – 0.8 mph and 11 + 0.8 mph = 10.2 mph and 11.8 mph. The interval (10.2, 11.8) is called the 95 % confidence interval. It tells us that the average drop in speed for the whole population is has a 95 % chance to be in this interval. The 0.8 mph is called the margin of error. Formula for Confidence Interval for Means In the previous example, we see that the confidence interval was constructed like this: Here 11 is the mean, , of the sample; 1.96 is the Z-value corresponding to 95% of the data; 2 is the standard deviation, σ, and the 22 is the sample size n. Thus, in general, the 95% confidence interval is The margin of error is 2 Spring 2010 Math 263 Deb Hughes Hallett Other Confidence Levels We have found a 95% confidence interval for the mean speed reduction for cars with radar detectors. It is also possible to estimate the mean speed reduction by using 90% and 99% confidence intervals from the same sample. Ex: How are the 95%, 90%, 99% confidence intervals related? Center of intervals: All centered at 11 mph Spread of intervals: The 90% confidence interval is shorter than the 95% confidence interval because the 90% interval does not have to be as sure that it contains the true value. The 99% confidence interval is longer than the 95% interval. Thus changing the confidence level makes the interval longer or shorter, but does not alter its center. Ex: Find Z-values for 90%, 95%, 99% confidence interval Confidence Level 90% 95% 99% z-values 1.645 1.96 2.575 Ex: What are the 90% and 99% confidence intervals for the drop in speed? 90% confidence: 99% confidence: Interpreting Confidence Intervals Informally we can say there’s a 90% chance that the mean speed drop is in the interval there’s a 95% chance that the mean speed drop is in the interval there’s a 99% chance that the mean speed drop is in the interval . However, this is not quite correct as the mean is a fixed number, so it either is, or isn’t, in these intervals—the probability is either 0 or 1. More properly, we say the method which produced a 95% interval covers the true mean 95% of the time. Ex: True or false: The 95% confidence interval tells us that 95% of the times we measure a speed drop, we will find it between 10.2 mph and 11.8 mph. False: The confidence interval tells us that the mean of the population is has a 95% chance of being in this interval, not that 95% of the individual readings are in this interval. 3 Spring 2010 Math 263 Deb Hughes Hallett Choosing Sample Size for the Margin of Error If the sample size was 50 (instead of 22), find the standard deviation of the sampling deviation of the sampling distribution, the margin of error and the 95% confidence interval. Standard deviation = mph Margin of Error = 1.96(0.28) = 0.55 mph Confidence Interval is: (11 – 0.55 mph, 11 + 0.55 mph) = (10.45 mph, 11.55 mph) Thus we can be 95% certain that the average drop in speed of the population of all cars with radar detectors is between 10.45 mph and 11.55 mph. Ex: Why does increasing the sample size decreases the margin of error? Explain mathematically and intuitively. Mathematically, the sample size is in the denominator of the expression for the standard deviation and the margin of error, so both decrease as the sample size increases. Intuitively, extreme values are more likely to average out in a larger sample, so the sampling distribution is less spread out––it has a smaller standard deviation. Thus the margin of error gets smaller as the sample size gets larger. Ex: If you needed a more precise estimate of the drop in speed to within 0.1 mph, how large a sample is required? We need the margin of error to be 0.1, and we solve for the sample size that achieves this. Since the margin of error , we have Thus a sample of 1537 cars is needed. 4 Spring 2010 Math 263 Deb Hughes Hallett Other Examples Ex: A US Department of Agriculture (USDA) study5 found that the mean price received by a sample of 22 farmers for corn was $2.08 per bushel with standard error $0.176 per bushel. Find a 95% confidence interval for the price of corn. What is the margin of error? We do not use the 22 as we are give that the standard error , so the confidence interval is (2.08 – 1.96(0.176), 2.08 + 1.96(0.176) = (1.74, 2.42) The true price was likely between $1.74 and $2.42. The margin of error is 1.96(0.176) = $0.345. Ex: The 95% confidence interval for the difference in birth weight6 (nonsmokers smokers) in grams for babies for mothers who do not smoke and those who do is (167, 595). Explain what this interval tells us. What is the best single number estimate of the weight difference? The study tells us that the weight difference for babies of smokers is estimated to be (167 + 595)/2 = 381 grams; the true value is likely to be between 167 and 595 grams.
Load more

Recommended publications
  • Measurement and Uncertainty Analysis Guide
    Measurements & Uncertainty Analysis Measurement and Uncertainty Analysis Guide “It is better to be roughly right than precisely wrong.” – Alan Greenspan Table of Contents THE UNCERTAINTY OF MEASUREMENTS .............................................................................................................. 2 RELATIVE (FRACTIONAL) UNCERTAINTY ............................................................................................................. 4 RELATIVE ERROR ....................................................................................................................................................... 5 TYPES OF UNCERTAINTY .......................................................................................................................................... 6 ESTIMATING EXPERIMENTAL UNCERTAINTY FOR A SINGLE MEASUREMENT ................................................ 9 ESTIMATING UNCERTAINTY IN REPEATED MEASUREMENTS ........................................................................... 9 STANDARD DEVIATION .......................................................................................................................................... 12 STANDARD DEVIATION OF THE MEAN (STANDARD ERROR) ......................................................................... 14 WHEN TO USE STANDARD DEVIATION VS STANDARD ERROR ...................................................................... 14 ANOMALOUS DATA ................................................................................................................................................
    [Show full text]
  • What Is This “Margin of Error”?
    What is this “Margin of Error”? On April 23, 2017, The Wall Street Journal reported: “Americans are dissatisfied with President Donald Trump as he nears his 100th day in office, with views of his effectiveness and ability to shake up Washington slipping, a new Wall Street Journal/NBC News poll finds. “More than half of Americans—some 54%—disapprove of the job Mr. Trump is doing as president, compared with 40% who approve, a 14‐point gap. That is a weaker showing than in the Journal/NBC News poll in late February, when disapproval outweighed approval by 4 points.” [Skipping to the end of the article …] “The Wall Street Journal/NBC News poll was based on nationwide telephone interviews with 900 adults from April 17‐20. It has a margin of error of plus or minus 3.27 percentage points, with larger margins of error for subgroups.” Throughout the modern world, every day brings news concerning the latest public opinion polls. At the end of each news report, you’ll be told the “margin of error” in the reported estimates. Every poll is, of course, subject to what’s called “sampling error”: Evenly if properly run, there’s a chance that the poll will, merely due to bad luck, end up with a randomly‐chosen sample of individuals which is not perfectly representative of the overall population. However, using the tools you learned in your “probability” course, we can measure the likelihood of such bad luck. Assume that the poll was not subject to any type of systematic bias (a critical assumption, unfortunately frequently not true in practice).
    [Show full text]
  • A Note on Confidence Interval Estimation and Margin of Error
    Journal of Statistics Education, Volume 18, Number 1 (2010) A Note on Confidence Interval Estimation and Margin of Error Dennis Gilliland Vince Melfi Michigan State University Journal of Statistics Education Volume 18, Number 1 (2010), www.amstat.org/publications/jse/v18n1/gilliland.pdf Copyright © 2010 by Dennis Gilliland and Vince Melfi all rights reserved. This text may be freely shared among individuals, but it may not be republished in any medium without express written consent from the authors and advance notification of the editor. Key Words: Confidence interval estimation; Margin of error; Interpretations; Misinterpretations. Abstract Confidence interval estimation is a fundamental technique in statistical inference. Margin of error is used to delimit the error in estimation. Dispelling misinterpretations that teachers and students give to these terms is important. In this note, we give examples of the confusion that can arise in regard to confidence interval estimation and margin of error. 1. Introduction Confidence interval estimation is a widely used method of inference and margin of error is a commonly used term, and these occupy a large part of introductory courses and textbooks in Statistics. It is well-known that these concepts are often misused and misunderstood. Examples of incorrect interpretations from a variety of sources, some “authoritative,” are given in Thornton and Thornton (2004). Recent doctoral theses of Liu (2005) and Noll (2007) address teacher and teaching assistant understanding of the concepts. Misunderstandings arise for a variety of reasons, some as simple as confusing population parameters and sample statistics. Some of the confusion in regard to the meaning and interpretation of these terms stems from the lack of appreciation of the difference between a random variable (a function) and its realization 1 Journal of Statistics Education, Volume 18, Number 1 (2010) (evaluation).
    [Show full text]
  • −1 ≤ R ≤ +1 FACT: −1 ≤ Ρ ≤ +1
    Ismor Fischer, 5/29/2012 7.2-1 7.2 Linear Correlation and Regression POPULATION Random Variables X, Y: numerical Definition: Population Linear Correlation Coefficient of X, Y σXY ρ = σX σY FACT: −1 ≤ ρ ≤ +1 SAMPLE, size n Definition: Sample Linear Correlation Coefficient of X, Y sxy ρˆ = r = sx sy 600 Example: r = = 0.907 strong, positive linear correlation 250 1750 FACT: −1 ≤ r ≤ +1 Any set of data points (xi, yi), i = 1, 2, …, n, having r > 0 (likewise, r < 0) is said to have a positive linear correlation (likewise, negative linear correlation). The linear correlation can be strong, moderate, or weak, depending on the magnitude. The closer r is to +1 (likewise, −1), the more strongly the points follow a straight line having some positive (likewise, negative) slope. The closer r is to 0, the weaker the linear correlation; if r = 0, then EITHER the points are uncorrelated (see 7.1), OR they are correlated, but nonlinearly (e.g., Y = X 2). Exercise: Draw a scatterplot of the following n = 7 data points, and compute r. (−3, 9), (−2, 4), (−1, 1), (0, 0), (1, 1), (2, 4), (3, 9) Ismor Fischer, 5/29/2012 7.2-2 s (Pearson’s) Sample Linear Correlation Coefficient r = xy sx sy uncorrelated r − 1 − 0.8 − 0.5 0 + 0.5 + 0.8 + 1 strong moderate weak moderate strong negative linear correlation positive linear correlation As X increases, Y decreases. As X increases, Y increases. As X decreases, Y increases. As X decreases, Y decreases. Some important exceptions to the “typical” cases above: r = 0, but X and Y are r > 0 in each of the two r > 0, only due to the effect of one correlated, nonlinearly individual subgroups, influential outlier; if removed, but r < 0 when combined then data are uncorrelated (r = 0) Ismor Fischer, 5/29/2012 7.2-3 Statistical Inference for ρ Suppose we now wish to conduct a formal test of… Hypothesis H0: ρ = 0 ⇔ “There is no linear correlation between X and Y.” vs.
    [Show full text]
  • Statistics Final Exam Review Notes How to Study for the Final?
    Statistics Final Exam Review Notes How to Study for the Final? • Study Exams • Study Final Exam Review Notes • Key Terms used in Stats Topic 1: Collecting Data And Bias Population: The collection of all people or objects you want to study. Census: Collecting data from every person or object in the population Sample: Collecting data from a subgroup of the population Random: Everyone in the population has an equal chance to be in the data Various ways of collecting data: • Convenience (asking friends and family) • Voluntary Response (putting a survey out into the world and allowing anyone to fill it out.) • Simple Random Sample (picking individual people or objects randomly usually with a random number generator) • Cluster (collecting data from groups of people in a population instead of one at a time, selecting classes and getting data from everyone in those classes) • Stratified (comparing two or more populations, so collecting sample data from each population, comparing a sample of women to a sample of men) • Systematic (getting data from every 20th person on a list or every 5th person that comes in a store) Bias: When a data set does not represent the population. Various kinds of bias • Sampling Bias (Not using randomization when you collect a sample, Voluntary Response, Convenience) • Response Bias (Controversial topics, people will not answer truthfully) • Non-Response Bias (people refuse to answer or take part in the data collecting) • Deliberate Bias (Deliberate lies about data, deliberately leave out certain groups of the population) • Question Bias (phrasing a question in a specific way in order to make people answer the way you want) The goal of data collecting is to get unbiased data that represents the population!! Topic 2: Experimental Design Related, Associated, Correlation ≠ Cause and Effect Why? Confounding Variables Confounding Variables: Variables that might influence the response variable (Y) other than the explanatory variable (X).
    [Show full text]
  • Chapter 10 Estimating with Confidence
    Chapter 10 Estimating with Confidence Vocabulary: Confidence interval significance level Margin of error Interval Confidence level Critical value A level C confidence interval test statistic z test statistic t distribution z distribution paired t procedures Test of significance Upper p critical value standard error Z* margin of error degrees of freedom Robust Calculator Skills: Z interval Z-test T interval 10-1 Confidence Intervals: The Basics (617-642) 1. In statistics, what is meant by a 95% confidence interval? 95% of samples will capture the true mean, parameter 2. Sketch and label a 95% confidence interval for the standard normal curve. 3. In a sampling distribution of x , why is the interval of numbers between x 2s called a 95% confidence interval? 95% of Normal population is within 2 standard deviations of the mean 4. Define a level C confidence interval. Estimate ± margin of error 5. Sketch and label a 90% confidence interval for the standard normal curve. 6. What does z* represent? The critical value of z or t. ( # of standard deviations) 7. What is the value of z* for a 90% confidence interval? Include a sketch. 1.645 8. What is the value of z* for a 95% confidence interval? Include a sketch. 1.96 9. What is the value of z* for a 99% confidence interval? Include a sketch. 2.576 10. What is meant by the upper p critical value of the standard normal distribution? Area to the right of a positive z* value 11. Explain how to find a level C confidence interval for an SRS of size n having unknown mean and known standard * deviation .
    [Show full text]
  • Determining Sample Size How to Ensure You Get the Correct Sample Size
    Determining Sample Size How to Ensure You Get the Correct Sample Size Scott M. Smith, Ph.D. qualtrics.com How many responses do you really need? This simple question is a never-ending quandary for researchers. A larger sample can yield more accurate results — but excessive responses can be pricey. Consequential research requires an understanding of the statistics that drive sample size decisions. A simple equation will help you put the migraine pills away and sample confidently. Before you can calculate a sample size, you need to determine a few things about the target population and the sample you need: 1. Population Size — How many total people fit your demographic? For instance, if you want to know about mothers living in the US, your population size would be the total number of mothers living in the US. Don’t worry if you are unsure about this number. It is common for the population to be unknown or approximated. 2. Margin of Error (Confidence Interval) — No sample will be perfect, so you need to decide how much error to allow. The confidence interval determines how much higher or lower than the population mean you are willing to let your sample mean fall. If you’ve ever seen a political poll on the news, you’ve seen a confidence interval. It will look something like this: “68% of voters said yes to Proposition Z, with a margin of error of +/- 5%.” 3. Confidence Level — How confident do you want to be that the actual mean falls within your confidence interval? The most common confidence intervals are 90% confident, 95% confident, and 99% confident.
    [Show full text]
  • Lecture Notes for Week 12
    Lecture 29 Nancy Pfenning Stats 1000 Reviewing Confidence Intervals and Tests for Ordinary One-Sample, Matched- Pairs, and Two-Sample Studies About Means Example Blood pressure X was measured for a sample of 10 black men. It was found that x¯ = 114:9, s = 10:84. Give a 90% confidence interval for mean blood pressure µ of all black men. [Note: we can assume that blood pressure tends to differ for different races or genders, and that is why a separate study is made of black men|the confounding variables of race and gender are being controlled.] This is an ordinary one-sample t procedure. s A level .90 confidence interval for µ is x¯ t∗ , where t∗ has 10 1 = 9 df. Consulting the pn − df = 9 row and .90 confidence column of Table A.2, we find t∗ = 1:83. Our confidence interval is 114:9 1:83 10:84 = (108:6; 121:2). p10 Here is what the MINITAB output looks like: N MEAN STDEV SE MEAN 90.0 PERCENT C.I. calcbeg 10 114.90 10.84 3.43 ( 108.62, 121.18) Example Blood pressure for a sample of 10 black men was measured at the beginning and end of a period of treatment with calcium supplements. To test at the 5% level if calcium was effective in lowering blood pressure, let the R.V. X denote decrease in blood pressure, beginning minus end, and µD would be the population mean decrease. This is a matched pairs procedure. To test H0 : µD = 0 vs. Ha : µD > 0, we find differences X to have sample mean d¯ = 5:0, ¯ d µ0 5 0 sample standard deviation s = 8:74.
    [Show full text]
  • The “Margin of Error” of Polls – Sampling Error, Bernoulli Processes, and Random Walks John Denker
    1 The \Margin of Error" of Polls { Sampling Error, Bernoulli Processes, and Random Walks John Denker We are often told that a poll of 1000 voters has a \margin of error" on the order of 4%. This is mostly nonsense. The statistical uncertainty (i.e. standard deviation) is at most 1.6% for any particular candidate, and is even less (about 0.3%) for candidates who are polling near 1% (or 99%). Even if you get the math right, it's still nonsense, since non-statistical uncertainties dominate. Contents 1 A Simple Three-Way Example2 2 How To Do It Wrong : NPR Example5 2.1 Original NPR Story.........................................5 3 Various Things that Can Go Wrong6 3.1 Voters are Not Coins.........................................6 3.2 Systematic Error...........................................7 3.3 Possible Origin of the Bogus «Sampling Error» Numbers.....................7 3.4 The Electoral College is a Noise Magnifier.............................8 4 A First Step in the Right Direction9 4.1 An Improved Bar Chart.......................................9 4.2 Mahalanobis Distance........................................ 10 5 Derivation of some Key Formulas 11 5.1 Multi-Dimensional Random Walk.................................. 11 5.2 Sampling and Polling......................................... 14 5.2.1 Example: 60:40 Coin Toss.................................. 15 5.2.2 Example: 49-49-2 Polling.................................. 15 5.3 Probability of Seeing Zero...................................... 16 5.4 Lopsided Error Bars......................................... 16 6 Philosophical and Pedagogical Remarks 19 7 Correlations and Covariance 20 8 References 21 1 A SIMPLE THREE-WAY EXAMPLE 2 1 A Simple Three-Way Example Suppose there are three candidates: Alice, Bob, and Carol. Suppose there are a huge number of voters { more than 100 million { and suppose (somewhat unrealistically) that they have firmly made up their minds.
    [Show full text]
  • 1 (Poisson) Model for (Sampling)Variability of Count in a Given Amount of “Experience” 1
    Course BIOS601: intensity rates:- models / inference / planning Contents1... 1 (Poisson) Model for (Sampling)Variability of Count in a given amount of “experience” 1. The Poisson Distribution • What it is, and some of its features The Poisson Distribution: what it is, and some of its features • How it arises, and derivations of its pdf • The (infinite number of) probabilities P0,P1, ..., Py, ..., of observing Y = • Examples of when it might apply 0, 1, 2, . , y, . “events” in a given amount of “experience.” • Examples of when it might not: • These probabilities, P rob[Y = y], or P [y]’s, or P ’s for short, are gov- “extra-” or “less-than-” Poisson variation Y y erned by a single parameter, the mean E[Y ] = µ. • Probability calculations y • P [y] = exp[−µ] µ /y! {note recurrence relation: Py = Py−1 × (µ/y).} 2. Inference re Poisson parameter (µ) • Shorthand: Y ∼ Poisson(µ). • First principles - exact and approximate -CIs 2 • V ar[Y ] = µ ; i.e., σY = µY . • SE-based CI’s 1/2 • Approximated by N(µ, σY = µ ) when µ >> 10. 3. Applications / worked examples • Open-ended (unlike Binomial), but in practice, has finite range. • Sample size for ‘counting statistics’ • Poisson data sometimes called ”numerator only”: (unlike Binomial) may • Headline: “Leukemia rate triples near Nuke Plant: Study” not “see” or count “non-events”: but there is (an invisible) denominator “behind’ the no. of incoming “wrong number” phone calls you receive. • Percutaneous Injuries in Medical Interns • Model-based Variance How it arises / derivations • From count (i.e., numerator) to Rate • Count of events (items) that occur randomly, with low homogeneous in- 4.
    [Show full text]
  • Overdispersed Models for Claim Count Distribution
    View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by DSpace at Tartu University Library TARTU UNIVERSITY FACULTY OF MATHEMATICS AND COMPUTER SCIENCE Institute of Mathematical Statistics Frazier Carsten Overdispersed Models for Claim Count Distribution Master’s Thesis Supervisor: Meelis K¨a¨arik, Ph.D TARTU 2013 Contents 1 Introduction 3 2 Classical Collective Risk Model 4 2.1 Properties ................................ 4 2.2 CompoundPoissonModel . 8 3 Compound Poisson Model with Different Insurance Periods 11 4 Overdispersed Models 14 4.1 Introduction............................... 14 4.2 CausesofOverdispersion. 15 4.3 Overdispersion in the Natural Exponential Family . .... 17 5 Handling Overdispersion in a More General Framework 22 5.1 MixedPoissonModel.......................... 23 5.2 NegativeBinomialModel. 24 6 Practical Applications of the Overdispersed Poisson Model 28 Kokkuv˜ote (eesti keeles) 38 References 40 Appendices 41 A Proofs 41 B Program codes 44 2 1 Introduction Constructing models to predict future loss events is a fundamental duty of actu- aries. However, large amounts of information are needed to derive such a model. When considering many similar data points (e.g., similar insurance policies or in- dividual claims), it is reasonable to create a collective risk model, which deals with all of these policies/claims together, rather than treating each one separately. By forming a collective risk model, it is possible to assess the expected activity of each individual policy. This information can then be used to calculate premiums (see, e.g., Gray & Pitts, 2012). There are several classical models that are commonly used to model the number of claims in a given time period.
    [Show full text]
  • Lecture Notes #7: Residual Analysis and Multiple Regression 7-1
    Lecture Notes #7: Residual Analysis and Multiple Regression 7-1 Richard Gonzalez Psych 613 Version 2.6 (Dec 2019) LECTURE NOTES #7: Residual Analysis and Multiple Regression Reading Assignment KNNL chapter 6 and chapter 10; CCWA chapters 4, 8, and 10 1. Statistical assumptions The standard regression model assumes that the residuals, or 's, are independently, identically distributed (usually called\iid"for short) as normal with µ = 0 and variance σ2. (a) Independence A residual should not be related to another residual. Situations where indepen- dence could be violated include repeated measures and time series because two or more residuals come from the same subject and hence may be correlated. An- other violation of independence comes from nested designs where subjects are clustered (such as in the same school, same family, same neighborhood). There are regression techniques that relax the independence assumption, as we saw in the repeated measures section of the course. (b) Identically distributed 2 As stated above, we assume that the residuals are distributed N(0, σ ). That is, we assume that each residual is sampled from the same normal distribution with a mean of zero and the same variance throughout. This is identical to the normality and equality of variance assumptions we had in the ANOVA. The terminology applies to regression in a slightly different manner, i.e., defined as constant variance along the entire range of the predictor variable, but the idea is the same. The MSE from the regression source table provides an estimate of the variance 2 σ for the 's. Usually, we don't have enough data at any given level of X to check whether the Y's are normally distributed with constant variance, so how should this Lecture Notes #7: Residual Analysis and Multiple Regression 7-2 assumption be checked? One may plot the residuals against the predicted scores (or instead the predictor variable).
    [Show full text]